International negotiations on a potential agreement to curb carbon emissions once the first commitment period of the Kyoto Protocol ends in 2012 are now entering a crucial phase. Over the next two years, the main characteristics of any climate agreement will need to be agreed upon, if an effective and affordable policy is to be achieved. Even if we leave aside political considerations, such a choice will not be easy. Global warming is long lasting, and thus any future policy should provide a long term-signal consistent with the ultimate objective of limiting temperature increase. This objective requires unprecedented action in terms of mobilising investment and resources in a variety of sectors, and may thus end up negatively affecting economic activities. It involves economic sectors as different as aluminium production and livestock management and calls for a concerted engagement of countries. It is likely that distributional consequences will give rise to free-riding incentives and may lead to the watering down of the stringency of any climate treaty. Policymakers involved in this process face a difficult task, in which prioritising is complicated. One way to reduce the dimensionality of the problem is to evaluate potential climate change agreements in an integrated fashion, through a set of quantitative indicators that make it possible to model at least part of this complexity.
A number of proposals for a post-Kyoto agreement have been put forward by scholars and policy experts in the past few years. Aldy and Stavins (2007) present a collection of architectures proposed within the Harvard Project on International Climate Agreements. In CEPR Discussion Paper 6995, we evaluate a number of these policy proposals using an integrated assessment model and thus providing one of the few evaluations that use a common and consistent framework of analysis.
Eight architectures for a post-2012 international climate agreement are assessed and compared along four relevant dimensions, namely their environmental effectiveness, economic efficiency, distributional implications, and potential enforceability. These indicators are meant to provide policymakers with a clearer picture of the main implications of some of the policy options currently on the climate negotiators’ tables. The eight policy architectures that have been analysed and compared are illustrated in Table 1.
Table 1. Architectures for agreement: Main features
|Name||Key Feature||Policy Instrument||Scope||Timing|
|R&D Coalition||R&D cooperation||R&D||Universal||Immediate|
|Dynamic Targets||Progressive efffort, particularly in DCs, to guarantee political acceptance. No long term stabilisation target||Cap and trade||Universal||Incremental|
Clubs of countries.
|Cap and trade and R&D||Partial||Incremental|
Delayed participation of DCs.
Final target is stabilisation at 550 CO2e
|Cap and trade||Universal||Immediate|
|Cap and trade with redistribution||Cap and trade per capita allocation to achieve 550 CO2e stabilisation target||Cap and trade||Partial||Incremental|
|Graduation||Bottom-up differentiated domestic targets to achieve 550 CO2e global stabilisation target||Cap and trade||Partial||Incremental|
|REDD||Inclusion of reducing emissions from deforestation in developing countries (REDD) in trading scheme to achieve 550 CO2e stabilisation target||Cap and trade||Universal||Immediate|
Global carbon tax
|Global tax to achieve 550 CO2e stabilisation target. Domestic recycling||Carbon tax||Universal||Immediate|
Universal agreements involve all regions, while partial agreements only require cooperation among a subset of regions. Agreements may require immediate abatement efforts from participating countries or they may take into account differences in countries’ ability to undertake abatement and, therefore, involve incremental participation, where some regions – usually transition economies and developing countries – are allowed to delay their participation and enter the agreement when they satisfy some agreed criteria. A further distinction across architectures is the type of policy instrument involved – most schemes adopt a cap and trade approach, but carbon taxes and R&D policies are also considered. Different criteria are also assumed for the allocation of emission allowances (top-down, such as equal per capita or historical emissions, as opposed to bottom-up, such as graduation).
Before looking at the performance of the above eight architectures for agreement, two key aspects should be pointed out. First, all proposed architectures focus on CO2 mitigation only, and do not include other greenhouse gases. Second, to prevent carbon leakage and free riding, all the proposed architectures envisage that countries at least commit to not exceeding their projected emissions under the business-as-usual (BAU) scenario, i.e. their projected emissions when no new climate policy is implemented. Countries are willing to commit to this minimum level of engagement, as it allows them to participate in the market for carbon permits, undertake cheaper abatement and receive financial benefits for it. The results of the comparison analysis are summarised in Table 2.
Table 2. Architectures for agreement: Main results
|Environmental Effectiveness (T°C above pre-industrial)||Economic Efficiency (npv GWP change wrt BAU, 5% d.r.)||Distributional impact (Gini 2100)||Enforceability (Countries with positive welfare change, out of 12)|
|Cap and trade with redistribution||2.76||-1.45%||0.198||3|
|Global carbon tax||2.76||-1.49%||0.178||0|
Significant differences across schemes are evident, driven by the focus of the architecture, highlighting the many facets of the problem. Some general indications do emerge. First of all, note that the architectures in Table 2 have been ordered by increasing environmental performance. Yet this is the same as ordering them by decreasing economic efficiency (the second column), i.e. increasing costs. It also corresponds to decreasing enforceability of the agreement. There is thus clear evidence of a trade-off between environmental effectiveness and economic efficiency and enforceability. In words, the higher the cost, the higher the environmental effectiveness of the policy architecture and the lower its enforceability. More ambitious policies are necessarily more costly.
Second, Table 2 reveals the difficulty of controlling global warming – none of the policy schemes can bring the 2100 temperature below a 2°C increase over pre-industrial levels. Most of them lead to a temperature increase closer to 3°C.
The inclusion of avoided deforestation (REDD) is shown to decrease the policy cost and thus to improve the enforceability of future agreements, as it provides additional incentives for participation for some developing countries. This result, based on the inclusion of avoided emissions from the Amazon forest in the carbon market, is likely to be amplified by the extension of REDD to other tropical forests countries such as Congo and Indonesia.
With regard to the distribution of economic benefits and costs of future agreements, all policies entail an improvement of income distribution across regions as measured by the Gini coefficient in 2100, though the effect is larger for graduation and climate clubs.
Finally, a policy aiming at R&D cooperation to stimulate the development and adoption of carbon free technologies, but without an explicit carbon price signal, is shown to have a positive effect on economic activity, and is thus likely to be the only one leading to a global, self-enforcing agreement. However, it is also likely to have a very limited effect on carbon concentrations, thus suggesting that R&D provisions may be a necessary but not sufficient element of an effective climate agreement.
Aldy, J., and Stavins, R. N. (Eds.) (2007), Architectures for Agreement: Addressing Global Climate Change in the Post-Kyoto World, Cambridge University Press, Cambridge, UK.
Bosetti, V., Carraro, C., Sgobbi, A., and Tavoni, M. (2008), "Modelling Economic Impacts of Alternative International Climate Policy Architectures: A Quantitative and Comparative Assessment of Architectures for Agreement," CEPR Discussion Paper 6995